Multiscale Simulation of Highly Heterogeneous and Fractured Reservoirs

During the last 2-3 years, SINTEF has in collaboration with NTNU developed new multiscale methods for computing pressure and flow velocities in petroleum reservoirs. In the current project we wish to develop these methods further and demonstrate to the oi l industry that multiscale methods can be used to compute flow patterns directly on field-scale geomodels of strongly heterogeneous and fractured reservoirs. In particular, we wish to demonstrate that the use of multiscale methodologies simplify and impro ve current workflows for geological and flow modelling. For SINTEF/NTNU it is crucial to establish an industrial "proof-of-concept" of the novel multiscale technology in order to acquire future research contracts. In the project we will use recent "mimet ic" discretisation principles to develop a new inhouse pressure solver that is robust with respect to anisotropy and nonorthogonal grids with curved faces. The new pressure solver will be used as a fine-scale solver in a multiscale mixed finite-element me thodology for efficient simulation of flow patterns in large and complex geomodels. To simulate fluid transport we will use two different approaches. 1) Develop fast direct solvers to be used if one is only interested in a single run with a large degree of sub-scale resolution. 2) Develop a novel multiscale solver as an alternative to use of pseudofunctions when one is interested in multiple runs with minor changes in the parameters. The pressure and transport solvers will be combined in an inhouse mult iscale simulator to be used to simulate flow on industry-standard geomodels consisting of multimillion cells or multiple geostatistical realisations to demonstrate a proof-of-concept to the industry.